<![CDATA[Discipline issues and the accuracy of attendance recording are crucial aspects of human resource management at PT Indomarco Prismatama, particularly in managing operational employees with shift work systems such as cashiers, warehouse staff, drivers, and administrative staff. The use of conventional attendance methods is highly vulnerable to data manipulation ("buddy punching") and administrative inefficiency. This research designs and develops a web-based fingerprint biometric attendance system using the HID DigitalPersona 4500 Fingerprint Reader sensor. The system development methodology employs the Waterfall model, consisting of requirement analysis, system design, coding, and testing phases. The uniqueness of this system lies in its hybrid architecture that integrates the main web data communication with a Python-based local service to execute fingerprint image feature extraction using the Scale-Invariant Feature Transform (SIFT) algorithm with a matching threshold of 0.28. System testing shows that the application can automatically verify employee identities in an average time of under 1.5 seconds without having to leave the main scanning page, and update attendance logs (check-in/check-out) in real-time on a MySQL database. The implementation results prove that this system successfully reduces the attendance fraud rate to 0% and accelerates the attendance data recapitulation process.]]>
Copyrights © 2026
